Keywords

1 Introduction

The term “diabetes insipidus” (DI) refers to the excretion of copious amounts of diluted urine (diabetes, flow; insipidus, tasteless), setting the disorder apart from diabetes mellitus (mellitus: sweet) and exemplifying the former practice of evaluating a differential diagnosis by tasting the patients’ urine. Polydipsia (excessive thirst) ensues as a consequence of polyuria. Under physiological circumstances, the antidiuretic hormone arginine vasopressin, AVP, is synthesized in specialized hypothalamic neurons that extend to the posterior pituitary, from where it is secreted when plasma osmolality rises or circulating blood volume falls. AVP promotes free water reabsorption from the tubular lumen across renal collecting duct cells into the interstitial and intravasal spaces (Fig. 14.1). If this antidiuretic axis is disturbed, DI results.

Fig. 14.1
figure 1

Schematic of the AVP-mediated antidiuretic mechanism. A polarized principal cell of the renal collecting duct is shown on the right. Circulating AVP binds to its renal receptor (V2R), a G protein-coupled seven transmembrane domain receptor, which is located at the basolateral membrane. Coupling of the receptor to the stimulatory G protein (Gs) activates adenylyl cyclase (AC), which in turn converts adenosine triphosphate (ATP) to cyclic adenosine monophosphate (cAMP). cAMP then activates the protein kinase A (PKA) pathway, which exerts two effects. PKA activates nuclear factors, which increases the transcription of the aquaporin 2 (AQP2) water channel gene. PKA phosphorylates AQP2, which leads to the formation of homotetramers (P-AQP2), the functional water channel unit, which is inserted into the apical membrane and allows free water flow from the collecting duct lumen into the cell. Water molecules exit the cell into the interstitial space through aquaporin 3 and 4 water channels (AQP3, AQP4), located at the basolateral membrane. Dephosphorylation of the P-AQP2 homotetramers is followed by their internalization and dissociation. [Reproduced, with permission, from Rutishauser et al. (2016)]

Full size image

Hereditary DI is a monogenetic disorder and usually follows a clear pattern of inheritance, reflecting either autosomal dominant, autosomal recessive, or X-linked transmission. Rarely, additional signs and symptoms are present. In these patients, DI thus constitutes a component of a syndromic disorder.

In this chapter, we will discuss the features of hereditary neurohypophyseal DI, focusing on both clinical and pathophysiological aspects.

2 Evolution of Research on Hereditary Neurohypophyseal DI

Hereditary DI has been described as early as 1841 (Lacombe 1841). In the late nineteenth and early twentieth century, large autosomal dominant pedigrees were reported in the German literature (Weil 1884, 1908; Gänsslen and Fritz 1924). While the molecular basis of the disorder was still unknown, important etiological insight came from post-mortem studies in patients with hereditary or “idiopathic” neurogenic DI (Gaupp 1941; Blotner 1958; Braverman et al. 1965; Green et al. 1967; Bergeron et al. 1991; Nagai et al. 1984). These investigations showed loss of vasopressinergic magnocellular neurons and mild gliosis in supraoptic and paraventricular hypothalamic nuclei, suggesting that neuronal damage had occurred specifically to AVP-producing cells and establishing the basis of the so-called neurotoxicity hypothesis. If a cell type-specific toxic effect, conferred e.g. by a mutant protein, could lead to cell death and eliminate AVP secretion, this would explain the autosomal dominant transmission observed in inherited neurogenic DI. In addition, a neurotoxic agent typically leads to progressive cell damage over a prolonged period of time, which would be in accordance with the clinical observation that the disease develops gradually after the initial manifestation several months to years after birth, albeit with high penetrance.

The gene encoding the precursor of AVP was an obvious candidate to cause hereditary neurogenic DI. The human genes for prepro-AVP and its homolog prepro-oxytocin (OT) were cloned over 30 years ago (Sausville et al. 1985). In 1991, the first paper reporting a heterozygous mutation in the AVP gene was published (Ito et al. 1991). Currently, 71 mutations are listed in the public domain of the Human Gene Mutation Database (HGMD, http://www.hgmd.cf.ac.uk).

Several animal models for neurohypophyseal DI have been described (Bernal et al. 2016). In the Brattleboro rat, studied since the 1960s (Valtin and Schroeder 1964), DI is transmitted in an autosomal recessive manner and results from a single base deletion in the AVP gene, which causes a frame shift and leads to inefficient translation of the mutant mRNA (Schmale et al. 1984). The first mouse model to reproduce many facets of human autosomal dominant neurohypophyseal DI was reported in 2003 (Russell et al. 2003). In these animals, a naturally occurring heterozygous human truncation mutation of the AVP precursor protein was introduced, resulting in a clear DI phenotype. Since then, additional rat (Si-Hoe et al. 2000; Davies and Murphy 2002; Castino et al. 2005) and mouse (Hayashi et al. 2009; Hagiwara et al. 2014) models have been published, with varying results regarding recapitulation of clinical polyuria/polydipsia and histological evidence of cytotoxicity in vasopressinergic neurons.

Since the 1990s, numerous expression studies have analysed intracellular trafficking of mutant vs. wild-type AVP hormone precursors (Olias et al. 1996; Beuret et al. 1999; Siggaard et al. 2005). In autosomal dominant neurogenic DI, retention of pathogenic AVP mutants in the endoplasmic reticulum (ER) has uniformly been found. Recent work has focused on aggregation of misfolded precursors in the ER and on ER-associated degradation, suggesting that insoluble fibrillar aggregates may contribute to or even cause cell death in dominant neurogenic DI (Rutishauser et al. 2016).

3 Differential Diagnosis of Hereditary Diabetes Insipidus (OMIM 192340; 125700)

In adults with ad libitum access to fluids, polyuria is defined as a 24-h urinary output of >45–50 ml/kg body weight (Robertson 2001). Polyuria should be quantified, as estimations by patients tend to be imprecise. Uncontrolled diabetes mellitus resulting in glucosuria and thus solute diuresis must be excluded. The differential diagnosis between neurohypophyseal DI, nephrogenic DI, and primary polydipsia can be facilitated by measuring copeptin concentrations at baseline and after sufficient stimulation (i.e. serum [Na] ≥ 150 mM) by water deprivation and/or hypertonic saline infusion (Timper et al. 2015; Fenske et al. 2018).

Although de novo mutations in previously unaffected families can occur (Rutishauser et al. 2002; Bourdet et al. 2016; Joshi et al. 2017), a positive family history of polyuria and polydipsia is usually present in hereditary forms (Babey et al. 2011). The mode of transmission and consideration of clinical characteristics often allow to make a presumptive diagnosis, but confirmatory genetic analysis is advocated for both nephrogenic and neurogenic familial DI (Bichet and Bockenhauer 2016; Rutishauser et al. 2016; Joshi et al. 2018). The majority of patients with hereditary nephrogenic DI have mutations in the vasopressin V2 receptor (AVPR2) gene, which is encoded on the X chromosome. Autosomal recessive or, less common, dominant transmission is associated with mutations in the gene encoding aquaporin 2 water channels. Hereditary nephrogenic DI, particularly the X-linked and recessive types, is present at birth and characterized by potentially life-threatening dehydration. Conversely, symptoms of familial neurohypophyseal DI typically develop gradually, manifesting months or even years after birth.

In hereditary neurohypophyseal DI, both autosomal recessive and, much more commonly, dominant transmission occur. Since these disorders have distinct pathophysiologies, they will be considered separately.

4 The Vasopressin Gene and Its Product

The homologous genes encoding prepro-vasopressin (AVP gene) and prepro-oxytocin (OT gene) are located in a head-to-head orientation on chromosome 20p13, separated by a ~12 kilobase intervening sequence. The AVP gene (Fig. 14.2) consists of three exons, encoding the 19-amino acid (aa) N-terminal signal peptide (SP), the AVP nonapeptide hormone, a 93-aa neurophysin II (NPII) moiety, and a C-terminal 39-aa glycoprotein. The latter, called copeptin, represents the main distinctive feature between the AVP and OT genes, as it is present only in the AVP precursor protein. After directing the nascent polypeptide chain to the ER, the SP is cleaved during cotranslational translocation. Prohormone folding in the ER implies AVP binding into a binding pocket of NPII (Wu et al. 2001; De Bree et al. 2003) and stabilization by eight disulfide bonds. After ER exit, the prohormone is transported through the Golgi apparatus to the trans-Golgi network, where it is sorted into neurosecretory granules and cleaved into AVP, NPII, and copeptin. In addition, AVP is C-terminally amidated. The three proteins are secreted into the circulation from nerve endings in the posterior pituitary by regulated exocytosis. Upon secretion, NPII and copeptin have no clinically appreciated function.

Fig. 14.2
figure 2

Domain organization of prepro-vasopressin neurophysin II. The hormone precursor consists of the 19-amino acid (aa) signal peptide (grey), the vasopressin nonapeptide (green), the 93-aa carrier protein neurophysin II, and the C-terminal glycopeptide, copeptin (yellow). The glycosylation site is indicated by a black diamond. There are a total of eight disulfide bridges, seven in the neurophysin II and one in the vasopressin moieties, respectively, indicated by red lines. Dots indicate heterozygous mutations associated with autosomal dominant neurohypophyseal diabetes insipidus; missense or deletion in black and nonsense or frameshift in pink. [Reproduced, with permission, from Beuret et al. (2017)]

Full size image

The majority of mutations in hereditary neurogenic DI are missense or nonsense point mutations, but deletions (Christensen et al. 2013; Luo et al. 2012) and indels have also been found (see http://www.hgmd.cf.ac.uk), as well as mutations altering splice sites (see below). No mutations have been described in the copeptin moiety.

5 Autosomal Recessive Neurohypophyseal DI

Compared to the much more common dominant type, this mode of inheritance in hereditary neurogenic DI is exceedingly rare. A recurrent recessive point mutation has been reported in two apparently unrelated Palestinian families originating from Texas (Willcutts et al. 1999) and Israel (Abu Libdeh et al. 2010). The mutation, passed on to the affected offspring by each of the heterozygous consanguineous parents, replaces proline at position 7 of the AVP moiety by leucin (P7L). Polyuria and polydipsia developed by the age of 15–24 months in affected children of one family and in the neonatal period in the other, but the reason for this discrepancy is unclear. Symptoms were successfully treated with desmopressin (DDAVP). At the cellular level, it was shown that the mutation caused a ~30-fold decrease in binding of AVP to its renal receptor. Conversely, when the P7L mutant was later expressed in cultured cells, its intracellular trafficking and secretion was normal (Birk et al. 2009).

Another child with very early-onset recessive neurogenic DI, born to consanguineous parents, was reported in 2013 (Christensen et al. 2013). Hypernatremia and urinary hypoosmolality were correctable by the administration of DDAVP, suggesting that the disorder was not caused by a defective AVPR2. Rather, a chromosomal deletion of ~10 kilobases in the AVP gene was identified on both alleles, assumedly abolishing gene transcription.

More recently, Canadian paediatricians reported on a child of Lebanese descent, born to asymptomatic parents and presenting with neurohypophyseal DI. Polyuria and polydipsia responded to DDAVP treatment, as did growth retardation. Sequence analysis of the AVP gene in the patient revealed a compound heterozygous state, with one allele carrying the known P7L mutation and the other a novel variant in the splice acceptor site of intron 1 (Bourdet et al. 2016). This case underscores the importance of genetic testing, even in the absence of a suggestive family history, in childhood-onset DI of uncertain origin.

6 Autosomal Dominant Neurohypophyseal DI

In the vast majority of published families with hereditary central DI, disease transmission follows a dominant pattern. Affected individuals gradually develop polyuria and polydipsia due to progressive AVP deficiency. Symptoms and signs typically start to manifest in infancy to early childhood. Disease penetrance is complete and symptoms are usually severe, although signal peptide mutations causing inefficient signal cleavage have been associated with later manifestation and a partial phenotype (McLeod et al. 1993; Siggaard et al. 2005; Toustrup et al. 2018). If children are deprived of adequate access to fluids, failure to thrive may occur, which is reversible by treatment with DDAVP (Brachet et al. 2011). Interestingly, the same mutation may result in variable disease severity, even among affected members of the same family (Repaske et al. 1997). Another peculiarity is partial recovery of polyuria in elderly males with familial neurohypophyseal DI (Rutishauser et al. 1996; Babey et al. 2011). The physiological basis of these observations is unclear.

6.1 ER Retention and Possible Links to Neuronal Cell Loss

Much insight has emerged from heterologous expression studies in cell culture experiments. The common denominator of all dominant neurogenic DI variants is that the mutant prohormone is retained in the ER by the ER quality control system, a machinery that ensures that only correctly folded proteins exit the ER (Araki and Nagata 2011). ER retention is common to the large group of “ER storage diseases”, such as cystic fibrosis (retention of the mutant CFTR chloride channel), LDL receptor defect, or nephrogenic DI (retention of mutant AVPR2 or aquaporin 2) (Rutishauser and Spiess 2002). However, as opposed to the majority of these disorders, in autosomal dominant neurohypophyseal DI, the disease does not result from the mere lack of functional protein, which should be compensated by the wild-type allele, but from a dominant-negative effect of the mutant over the wild-type counterpart.

ER retention can be demonstrated by pulse-chase and immunoprecipitation experiments using cells expressing mutant or wild-type (control) provasopressin (Fig. 14.3). ER-retained proteins produce a typical reticular or more densely packed aggregate pattern when visualized by immunofluorescence, co-localizing with ER-resident marker proteins (Fig. 14.4).

Fig. 14.3
figure 3

ER retention of a dominant DI mutation. Wild-type vasopressin precursor (V) and the pathogenic mutant ΔG227 [Vm; see Rutishauser et al. (1996)] were expressed in transiently transfected COS-7 fibroblast cells. After pulse labelling for 30 min with [35S]methionine, cells were chased for the indicated durations with excess unlabelled methionine. Media (M) and cell lysates (C) were immunoprecipitated and subjected to gel electrophoresis and autoradiography. The wild-type protein was rapidly secreted into the medium (lanes 1–8), while the mutant was retained quantitatively in the cells even after a chase of 4 h (lanes 13–22). Some samples were incubated with (+) or without (−) endoglycosidase H. Cell lysates (lanes 11, 12 and 25, 26) were sensitive to this treatment, indicating high-mannose glycosylation typical for the ER. In contrast, wild-type protein found in the media (lanes 9, 10) was endoglycosidase H-resistant, indicating that it had adopted complex glycosylation in the Golgi apparatus. (Reproduced with permission. This research was originally published in the Journal of Biological Chemistry. Beuret N., Rutishauser J., Bider M.D., Spiess M. Mechanism of endoplasmic reticulum retention of mutant vasopressin precursor caused by a signal peptide truncation associated with diabetes insipidus. J. Biol. Chem. 1999; 274(27): 18965–18972. © The American Society for Biochemistry and Molecular Biology)

Full size image
Fig. 14.4
figure 4

ER localization of a dominant DI mutant by immunofluorescence. COS-1 fibroblast cells were transiently transfected with the dominant mutant C28Y. Cells were stained with antibodies against NP II (left panel) or the ER-resident protein BAP31 (right panel). The mutant protein produces a reticular staining pattern typical for the ER (NPII staining, right cell). Some cells show a more aggregate pattern (NPII staining, left cell). Anti-BAP31 staining shows co-localization with the C28Y mutant. Size bar: 20 μm [Adapted, with permission, from Rutishauser et al. (2016)]

Full size image

Disruption of ER homeostasis, i.e. the mismatch between the load of proteins populating the ER and the ER folding and secretory capacity, activates a program of pathways signalling from the ER lumen to the cytoplasm and the nucleus (Chapman et al. 1998; Mori 2000). This program, termed ER stress response or unfolded protein response (UPR), is made up of several distinct components; e.g. upregulation of ER-resident chaperones and folding enzymes occurs, or programmed cell death pathways may become activated (Oyadomari and Mori 2004). In autosomal dominant neurohypophyseal DI, while the mechanism of neurotoxicity appears to be different from apoptosis (Ito and Jameson 1997), a heterozygous knock-in mouse model, carrying a human truncation provasopressin mutant, reproduced not only the clinical phenotype of AVP-deficient DI, but also demonstrated by immunostaining an apparent reduction in vasopressinergic neurons in the hypothalamus (Russell et al. 2003). This seemingly direct evidence for the neurotoxicity hypothesis has been questioned, however, in later experiments with mice carrying the same heterozygous truncation mutant. By in situ hybridization for AVP mRNA, vasopressinergic neurons were detectable despite the presence of polyuria in the mutant mice (Hayashi et al. 2009). The polyuria phenotype thus preceded neuronal loss. In the ER and nuclei of vasopressinergic cells, large aggregations were detected which contained markers of autophagy, a lysosomal clearance mechanism of intracellular protein waste (Hagiwara et al. 2014). Thus, in the pathophysiology of autosomal dominant neurogenic DI, the autophagy pathway may be involved, but the exact nature of the link between ER retention and toxic impact on neurons remains unclear.

6.2 Degradation and Fibrillar Aggregation of Pathogenic Provasopressin Mutants

ER-retained misfolded proteins are retrotranslocated from the ER lumen into the cytosol and targeted for ER-associated degradation (ERAD) by the proteasome (Ciechanover and Schwartz 1998; Hershko et al. 2000). This process has been demonstrated in the case of pathogenic provasopressin mutants in cell culture experiments where proteasome inhibitors were used (Friberg et al. 2004) (Fig. 14.5). Interestingly, it was found that not only mutant but also a portion of wild-type precursors were degraded by the proteasome pathway. Indirect evidence for a “system overload” came from the early observation that a dominant AVP mutant associated with severe DI formed disulfide-linked oligomers (Beuret et al. 1999). It was later confirmed that these homo-oligomers accumulated in cells transfected with pathogenic AVP mutants (Birk et al. 2009). Further analysis of transfected cells by immunofluorescence staining and immunogold electron microscopy revealed that large aggregations with a fibrillar substructure gradually accumulated in the ER lumen (Birk et al. 2009; Beuret et al. 2017) (Fig. 14.6). These findings indicated that autosomal dominant neurohypophyseal DI shares essential characteristics with neurodegenerative disorders such as Huntington’s, Alzheimer’s, or Parkinson’s disease associated with fibrillar (amyloid) aggregation.

Fig. 14.5
figure 5

Degradation of provasopressin by the proteasome. Wild-type (wt) and three different dominant DI mutants were expressed in transiently transfected COS-1 cells. G57S and ΔE47 are two mutations in the NPII moiety of the precursor. The signal peptide mutant A(−1)T causes partial signal cleavage deficiency. Cells were pulse-labelled with [35S]methionine and chased for 0 or 6 h in the absence (control; lanes 1–8) or presence (lanes 9–16) of proteasomal inhibitors, 250 μM ALLN (panel a) or 25 μM lactacystin (panel b). Cell lysates (panel a, top) and media (panel a, bottom) were subjected to immunoprecipitation, gel electrophoresis, and autoradiography. After labelling, the precursor was detected as a major cytosolic ~21 kDa species, corresponding to glycosylated provasopressin. In the case of the A(−1)T mutant, a slightly larger band was also visible, representing uncleaved prepro-vasopressin. After 6 h of chase, wild-type and the signal-cleaved fraction of the A(−1)T mutant were detected in the media. Hardly any secretion-deficient mutants were detectable in the cells after 6 h, suggesting that they had been almost completely degraded. The addition of the proteasome inhibitors ALLN and lactacystin, but not the lysosomal inhibitors leupeptin and pepstatin (not shown), stabilized the cytosolic 21 kDa mutant and wild-type proteins. In addition to the full-length precursor, three cytosolic degradation intermediates of ~17–19 kDa [asterisks; panel a top and panel b] were also produced after labelling. These bands were equally stabilized by ALLN and lactacystin. These results indicate that retained dominant DI mutants—and to a smaller degree also wild-type provasopressin—is degraded by proteasomes. (Reproduced with permission. This research was originally published in the Journal of Biological Chemistry. Friberg M.A., Spiess M., Rutishauser J. Degradation of Wild-type Vasopressin Precursor and Pathogenic Mutants by the Proteasome. J. Biol. Chem. 2004; 279(29): 19441–19447. © The American Society for Biochemistry and Molecular Biology)

Full size image
Fig. 14.6
figure 6

ER aggregation of a C-terminally truncated DI mutant. In the shortest known naturally occurring dominant DI mutant, C61X, the codon for cysteine at position 61 is altered to a stop codon. This truncation mutant is not recognized by anti-provasopressin antibody raised against the C-terminal portion of NPII and copeptin and was myc-tagged for detection by anti-myc antibody. The C61myc mutant was expressed in transiently transfected mouse hippocampal HN10 neuroblastoma cells and analysed by immunofluorescence microscopy after co-staining for myc and the ER chaperone calreticulin (panel a, top and middle). Nuclei were stained with DAPI (blue). The two proteins co-localized in aggregations (panel a, bottom), indicating that the mutant was retained in the ER. Immunogold electron microscopy demonstrated that the provasopressin aggregates consisted of a network of fibrils, decorated by gold particles seen as black dots (panel b, top and bottom right). To enhance clarity, the fibrils were manually highlighted and the black dots coloured in red (panel b, bottom left). Size bar: 200 nm. [Reproduced, with permission, from Beuret et al. (2017)]

Full size image

Recent studies have shown that two sequences within the provasopressin prohormone independently confer the ability to form fibrils, namely, the AVP nonapeptide and the C-terminal copeptin (Beuret et al. 2017). Interestingly, the same two sequences also mediate physiological aggregation and sorting of the prohormone into neurosecretory granules (Beuret et al. 2017), lending support to the proposal that secretory granules are “functional amyloids” (Maji et al. 2009).

The current hypothesis thus states that the amount of mutant AVP precursor molecules exceeds the capacity of the degradation machinery, leading to accumulation and deposition of large fibrillar aggregations in the ER and, potentially, to neuronal damage by undefined mechanisms. Recent experimental work has addressed the role of the ER quality control and degradation system in the context of vasopressin physiology. Remarkably, mice with an intact AVP gene, but deficient in Sel1L protein, a component of the ERAD complex, develop AVP-deficient DI without showing evidence of damage in magnocellular neurons (Shi et al. 2017). In a mouse neuroblastoma cell line deficient in Sel1L, wild-type and mutant vasopressin precursors accumulate in amyloid-like fibrillar aggregations (Shi et al. 2017). These data show that a functional ERAD is essential for maintaining the antidiuretic system in the wild-type context already and support the pathophysiological concepts outlined above which underlie autosomal dominant neurohypophyseal DI.

7 X-Linked Neurohypophyseal DI

Only one family tree with apparent X-linked transmission of neurogenic DI has been reported in abstract form (Habiby et al. 1996). Females are unaffected, and males show desmopressin-responsive DI of varying severity, manifesting months to years after birth (Babey et al. 2011). The phenotype has been linked to a region on chromosome Xq28. Importantly, both the AVP and AVPR2 genes in affected males are normal. The gene causing this form of hereditary DI has not been identified so far.

8 Desmopressin-Responsive Diabetes Insipidus in the Context of the Wolfram Syndrome 1 (DIDMOAD Syndrome)

Patients affected with this rare autosomal recessive syndrome, first described in 1938 (Wolfram and Wagener 1938), carry homozygous or compound heterozygous mutations in the WFS1 gene (Inoue et al. 1998; Strom et al. 1998) (OMIM 222300). Diabetes insipidus, diabetes mellitus, optic atrophy, and sensorineural deafness (DIDMOAD) are the components of the complex disorder, with DI manifesting in up to 70% of affected individuals (Barrett et al. 1995). Insulin-dependent diabetes mellitus and bilateral optic atrophy are the two features requested to make a diagnosis, but besides DI and deafness, psychiatric illness associated with a high risk of suicide (Swift et al. 1991, 1998) and brain stem and cerebellar disease (Chaussenot et al. 2011) have also been observed. Prognosis is poor when patients develop neurological dysfunction.

The WFS1 gene, located in chromosome 4p16.1, encodes wolframin, an ER-resident transmembrane protein which is believed to control ER calcium levels and attenuate ER stress in pancreatic beta cells (Fonseca et al. 2005; Rigoli et al. 2018). Heterozygous mutations in WFS1, resulting in deficiency in wolframin, may trigger the ER stress response and eventually activation of pro-apoptotic pathways (Delprat et al. 2018).

9 Conclusions

The clinical presentation, along with a careful medical history, will often allow a correct differential diagnosis in hereditary DI, which is confirmed by functional tests and genetic analysis. The neurogenic form is most often transmitted in an autosomal dominant fashion, and symptoms develop gradually, rarely endangering affected individuals. In abrupt neonatal onset, much more often caused by nephrogenic than neurogenic DI, rapid hydration treatment is key. Mutational analysis is recommended in all patients with suspected hereditary DI; importantly, a genetic basis should be considered in idiopathic childhood DI even in the absence of a family history, because de novo mutations in the responsible gene may occur.